It is difficult to create a definitive pro/cons sheet because a lot depends on the virtualization infrastructure and ops of the customer.
All PAM versions.
We have several large customers who use hardware for the Primary and Secondary Site Clusters but use OVA for DR site. These customers want strict control over the management of the application and having dedicated hardware provides this. Often organizations have a separate group responsible for optimizing hypervisor utilization which means moving the workload around. We do provide guidance on this in the docs.
Cluster Maintenance (broadcom.com)
Snap-shots and changing the available resources to the application are generally the upside of virtualization but again it depends on the quality of the bare metal hypervisor.
For hardware the downsides are; the 3 years cycle to replace it, approval for racking in the data centers which may be remote and potential for hardware failures. However, for the latter, in general, this has not been an issue (mostly just customers bricking a device by poor reboot practices). The hardware we currently ship has 64GB RAM, 1273v6 Intel Xeon processor with dual 240GB SSD units and dual PSU. Compared to the 304 hardware NYCB currently have, the difference is mainly the newer CPU and larger (4X) memory.
A hybrid cluster of hardware, OVA and cloud (AWS/Azure) is also possible and we have customers with such a deployment. So in reality it depends on the customer's desired architecture.